/*
- * Copyright (c) 200
6-2009, Salvatore Sanfilippo <antirez at gmail dot com>
+ * Copyright (c) 200
9-2010, Salvatore Sanfilippo <antirez at gmail dot com>
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* POSSIBILITY OF SUCH DAMAGE.
*/
-#define REDIS_VERSION "1.3.3"
+#define REDIS_VERSION "1.3.4"
#include "fmacros.h"
#include "config.h"
#include <time.h>
#include <unistd.h>
#define __USE_POSIX199309
+#define __USE_UNIX98
#include <signal.h>
#ifdef HAVE_BACKTRACE
#include "zmalloc.h" /* total memory usage aware version of malloc/free */
#include "lzf.h" /* LZF compression library */
#include "pqsort.h" /* Partial qsort for SORT+LIMIT */
+#include "zipmap.h"
/* Error codes */
#define REDIS_OK 0
#define REDIS_ZSET 3
#define REDIS_HASH 4
-/* Objects encoding */
+/* Objects encoding. Some kind of objects like Strings and Hashes can be
+ * internally represented in multiple ways. The 'encoding' field of the object
+ * is set to one of this fields for this object. */
#define REDIS_ENCODING_RAW 0 /* Raw representation */
#define REDIS_ENCODING_INT 1 /* Encoded as integer */
+#define REDIS_ENCODING_ZIPMAP 2 /* Encoded as zipmap */
+#define REDIS_ENCODING_HT 3 /* Encoded as an hash table */
+
+static char* strencoding[] = {
+ "raw", "int", "zipmap", "hashtable"
+};
/* Object types only used for dumping to disk */
#define REDIS_EXPIRETIME 253
#define APPENDFSYNC_ALWAYS 1
#define APPENDFSYNC_EVERYSEC 2
+/* Hashes related defaults */
+#define REDIS_HASH_MAX_ZIPMAP_ENTRIES 64
+#define REDIS_HASH_MAX_ZIPMAP_VALUE 512
+
/* We can print the stacktrace, so our assert is defined this way: */
#define redisAssert(_e) ((_e)?(void)0 : (_redisAssert(#_e,__FILE__,__LINE__),_exit(1)))
static void _redisAssert(char *estr, char *file, int line);
off_t vm_page_size;
off_t vm_pages;
unsigned long long vm_max_memory;
+ /* Hashes config */
+ size_t hash_max_zipmap_entries;
+ size_t hash_max_zipmap_value;
/* Virtual memory state */
FILE *vm_fp;
int vm_fd;
redisCommandProc *proc;
int arity;
int flags;
+ /* What keys should be loaded in background when calling this command? */
+ int vm_firstkey; /* The first argument that's a key (0 = no keys) */
+ int vm_lastkey; /* THe last argument that's a key */
+ int vm_keystep; /* The step between first and last key */
};
struct redisFunctionSym {
typedef struct zskiplistNode {
struct zskiplistNode **forward;
struct zskiplistNode *backward;
+ unsigned int *span;
double score;
robj *obj;
} zskiplistNode;
static struct redisCommand *lookupCommand(char *name);
static void call(redisClient *c, struct redisCommand *cmd);
static void resetClient(redisClient *c);
+static void convertToRealHash(robj *o);
static void authCommand(redisClient *c);
static void pingCommand(redisClient *c);
static void zincrbyCommand(redisClient *c);
static void zrangeCommand(redisClient *c);
static void zrangebyscoreCommand(redisClient *c);
+static void zcountCommand(redisClient *c);
static void zrevrangeCommand(redisClient *c);
static void zcardCommand(redisClient *c);
static void zremCommand(redisClient *c);
static void zremrangebyscoreCommand(redisClient *c);
static void multiCommand(redisClient *c);
static void execCommand(redisClient *c);
+static void discardCommand(redisClient *c);
static void blpopCommand(redisClient *c);
static void brpopCommand(redisClient *c);
static void appendCommand(redisClient *c);
+static void substrCommand(redisClient *c);
+static void zrankCommand(redisClient *c);
+static void zrevrankCommand(redisClient *c);
+static void hsetCommand(redisClient *c);
+static void hgetCommand(redisClient *c);
+static void hdelCommand(redisClient *c);
+static void zremrangebyrankCommand(redisClient *c);
+static void zunionCommand(redisClient *c);
+static void zinterCommand(redisClient *c);
/*================================= Globals ================================= */
/* Global vars */
static struct redisServer server; /* server global state */
static struct redisCommand cmdTable[] = {
- {"get",getCommand,2,REDIS_CMD_INLINE},
- {"set",setCommand,3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM},
- {"setnx",setnxCommand,3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM},
- {"append",appendCommand,3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM},
- {"del",delCommand,-2,REDIS_CMD_INLINE},
- {"exists",existsCommand,2,REDIS_CMD_INLINE},
- {"incr",incrCommand,2,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM},
- {"decr",decrCommand,2,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM},
- {"mget",mgetCommand,-2,REDIS_CMD_INLINE},
- {"rpush",rpushCommand,3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM},
- {"lpush",lpushCommand,3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM},
- {"rpop",rpopCommand,2,REDIS_CMD_INLINE},
- {"lpop",lpopCommand,2,REDIS_CMD_INLINE},
- {"brpop",brpopCommand,-3,REDIS_CMD_INLINE},
- {"blpop",blpopCommand,-3,REDIS_CMD_INLINE},
- {"llen",llenCommand,2,REDIS_CMD_INLINE},
- {"lindex",lindexCommand,3,REDIS_CMD_INLINE},
- {"lset",lsetCommand,4,REDIS_CMD_BULK|REDIS_CMD_DENYOOM},
- {"lrange",lrangeCommand,4,REDIS_CMD_INLINE},
- {"ltrim",ltrimCommand,4,REDIS_CMD_INLINE},
- {"lrem",lremCommand,4,REDIS_CMD_BULK},
- {"rpoplpush",rpoplpushcommand,3,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM},
- {"sadd",saddCommand,3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM},
- {"srem",sremCommand,3,REDIS_CMD_BULK},
- {"smove",smoveCommand,4,REDIS_CMD_BULK},
- {"sismember",sismemberCommand,3,REDIS_CMD_BULK},
- {"scard",scardCommand,2,REDIS_CMD_INLINE},
- {"spop",spopCommand,2,REDIS_CMD_INLINE},
- {"srandmember",srandmemberCommand,2,REDIS_CMD_INLINE},
- {"sinter",sinterCommand,-2,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM},
- {"sinterstore",sinterstoreCommand,-3,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM},
- {"sunion",sunionCommand,-2,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM},
- {"sunionstore",sunionstoreCommand,-3,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM},
- {"sdiff",sdiffCommand,-2,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM},
- {"sdiffstore",sdiffstoreCommand,-3,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM},
- {"smembers",sinterCommand,2,REDIS_CMD_INLINE},
- {"zadd",zaddCommand,4,REDIS_CMD_BULK|REDIS_CMD_DENYOOM},
- {"zincrby",zincrbyCommand,4,REDIS_CMD_BULK|REDIS_CMD_DENYOOM},
- {"zrem",zremCommand,3,REDIS_CMD_BULK},
- {"zremrangebyscore",zremrangebyscoreCommand,4,REDIS_CMD_INLINE},
- {"zrange",zrangeCommand,-4,REDIS_CMD_INLINE},
- {"zrangebyscore",zrangebyscoreCommand,-4,REDIS_CMD_INLINE},
- {"zrevrange",zrevrangeCommand,-4,REDIS_CMD_INLINE},
- {"zcard",zcardCommand,2,REDIS_CMD_INLINE},
- {"zscore",zscoreCommand,3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM},
- {"incrby",incrbyCommand,3,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM},
- {"decrby",decrbyCommand,3,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM},
- {"getset",getsetCommand,3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM},
- {"mset",msetCommand,-3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM},
- {"msetnx",msetnxCommand,-3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM},
- {"randomkey",randomkeyCommand,1,REDIS_CMD_INLINE},
- {"select",selectCommand,2,REDIS_CMD_INLINE},
- {"move",moveCommand,3,REDIS_CMD_INLINE},
- {"rename",renameCommand,3,REDIS_CMD_INLINE},
- {"renamenx",renamenxCommand,3,REDIS_CMD_INLINE},
- {"expire",expireCommand,3,REDIS_CMD_INLINE},
- {"expireat",expireatCommand,3,REDIS_CMD_INLINE},
- {"keys",keysCommand,2,REDIS_CMD_INLINE},
- {"dbsize",dbsizeCommand,1,REDIS_CMD_INLINE},
- {"auth",authCommand,2,REDIS_CMD_INLINE},
- {"ping",pingCommand,1,REDIS_CMD_INLINE},
- {"echo",echoCommand,2,REDIS_CMD_BULK},
- {"save",saveCommand,1,REDIS_CMD_INLINE},
- {"bgsave",bgsaveCommand,1,REDIS_CMD_INLINE},
- {"bgrewriteaof",bgrewriteaofCommand,1,REDIS_CMD_INLINE},
- {"shutdown",shutdownCommand,1,REDIS_CMD_INLINE},
- {"lastsave",lastsaveCommand,1,REDIS_CMD_INLINE},
- {"type",typeCommand,2,REDIS_CMD_INLINE},
- {"multi",multiCommand,1,REDIS_CMD_INLINE},
- {"exec",execCommand,1,REDIS_CMD_INLINE},
- {"sync",syncCommand,1,REDIS_CMD_INLINE},
- {"flushdb",flushdbCommand,1,REDIS_CMD_INLINE},
- {"flushall",flushallCommand,1,REDIS_CMD_INLINE},
- {"sort",sortCommand,-2,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM},
- {"info",infoCommand,1,REDIS_CMD_INLINE},
- {"monitor",monitorCommand,1,REDIS_CMD_INLINE},
- {"ttl",ttlCommand,2,REDIS_CMD_INLINE},
- {"slaveof",slaveofCommand,3,REDIS_CMD_INLINE},
- {"debug",debugCommand,-2,REDIS_CMD_INLINE},
- {NULL,NULL,0,0}
+ {"get",getCommand,2,REDIS_CMD_INLINE,1,1,1},
+ {"set",setCommand,3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,0,0,0},
+ {"setnx",setnxCommand,3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,0,0,0},
+ {"append",appendCommand,3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,1,1,1},
+ {"substr",substrCommand,4,REDIS_CMD_INLINE,1,1,1},
+ {"del",delCommand,-2,REDIS_CMD_INLINE,0,0,0},
+ {"exists",existsCommand,2,REDIS_CMD_INLINE,1,1,1},
+ {"incr",incrCommand,2,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM,1,1,1},
+ {"decr",decrCommand,2,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM,1,1,1},
+ {"mget",mgetCommand,-2,REDIS_CMD_INLINE,1,-1,1},
+ {"rpush",rpushCommand,3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,1,1,1},
+ {"lpush",lpushCommand,3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,1,1,1},
+ {"rpop",rpopCommand,2,REDIS_CMD_INLINE,1,1,1},
+ {"lpop",lpopCommand,2,REDIS_CMD_INLINE,1,1,1},
+ {"brpop",brpopCommand,-3,REDIS_CMD_INLINE,1,1,1},
+ {"blpop",blpopCommand,-3,REDIS_CMD_INLINE,1,1,1},
+ {"llen",llenCommand,2,REDIS_CMD_INLINE,1,1,1},
+ {"lindex",lindexCommand,3,REDIS_CMD_INLINE,1,1,1},
+ {"lset",lsetCommand,4,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,1,1,1},
+ {"lrange",lrangeCommand,4,REDIS_CMD_INLINE,1,1,1},
+ {"ltrim",ltrimCommand,4,REDIS_CMD_INLINE,1,1,1},
+ {"lrem",lremCommand,4,REDIS_CMD_BULK,1,1,1},
+ {"rpoplpush",rpoplpushcommand,3,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM,1,2,1},
+ {"sadd",saddCommand,3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,1,1,1},
+ {"srem",sremCommand,3,REDIS_CMD_BULK,1,1,1},
+ {"smove",smoveCommand,4,REDIS_CMD_BULK,1,2,1},
+ {"sismember",sismemberCommand,3,REDIS_CMD_BULK,1,1,1},
+ {"scard",scardCommand,2,REDIS_CMD_INLINE,1,1,1},
+ {"spop",spopCommand,2,REDIS_CMD_INLINE,1,1,1},
+ {"srandmember",srandmemberCommand,2,REDIS_CMD_INLINE,1,1,1},
+ {"sinter",sinterCommand,-2,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM,1,-1,1},
+ {"sinterstore",sinterstoreCommand,-3,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM,2,-1,1},
+ {"sunion",sunionCommand,-2,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM,1,-1,1},
+ {"sunionstore",sunionstoreCommand,-3,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM,2,-1,1},
+ {"sdiff",sdiffCommand,-2,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM,1,-1,1},
+ {"sdiffstore",sdiffstoreCommand,-3,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM,2,-1,1},
+ {"smembers",sinterCommand,2,REDIS_CMD_INLINE,1,1,1},
+ {"zadd",zaddCommand,4,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,1,1,1},
+ {"zincrby",zincrbyCommand,4,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,1,1,1},
+ {"zrem",zremCommand,3,REDIS_CMD_BULK,1,1,1},
+ {"zremrangebyscore",zremrangebyscoreCommand,4,REDIS_CMD_INLINE,1,1,1},
+ {"zremrangebyrank",zremrangebyrankCommand,4,REDIS_CMD_INLINE,1,1,1},
+ {"zunion",zunionCommand,-4,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM,0,0,0},
+ {"zinter",zinterCommand,-4,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM,0,0,0},
+ {"zrange",zrangeCommand,-4,REDIS_CMD_INLINE,1,1,1},
+ {"zrangebyscore",zrangebyscoreCommand,-4,REDIS_CMD_INLINE,1,1,1},
+ {"zcount",zcountCommand,4,REDIS_CMD_INLINE,1,1,1},
+ {"zrevrange",zrevrangeCommand,-4,REDIS_CMD_INLINE,1,1,1},
+ {"zcard",zcardCommand,2,REDIS_CMD_INLINE,1,1,1},
+ {"zscore",zscoreCommand,3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,1,1,1},
+ {"zrank",zrankCommand,3,REDIS_CMD_BULK,1,1,1},
+ {"zrevrank",zrevrankCommand,3,REDIS_CMD_BULK,1,1,1},
+ {"hset",hsetCommand,4,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,1,1,1},
+ {"hget",hgetCommand,3,REDIS_CMD_BULK,1,1,1},
+ {"hdel",hdelCommand,3,REDIS_CMD_BULK,1,1,1},
+ {"incrby",incrbyCommand,3,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM,1,1,1},
+ {"decrby",decrbyCommand,3,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM,1,1,1},
+ {"getset",getsetCommand,3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,1,1,1},
+ {"mset",msetCommand,-3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,1,-1,2},
+ {"msetnx",msetnxCommand,-3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,1,-1,2},
+ {"randomkey",randomkeyCommand,1,REDIS_CMD_INLINE,0,0,0},
+ {"select",selectCommand,2,REDIS_CMD_INLINE,0,0,0},
+ {"move",moveCommand,3,REDIS_CMD_INLINE,1,1,1},
+ {"rename",renameCommand,3,REDIS_CMD_INLINE,1,1,1},
+ {"renamenx",renamenxCommand,3,REDIS_CMD_INLINE,1,1,1},
+ {"expire",expireCommand,3,REDIS_CMD_INLINE,0,0,0},
+ {"expireat",expireatCommand,3,REDIS_CMD_INLINE,0,0,0},
+ {"keys",keysCommand,2,REDIS_CMD_INLINE,0,0,0},
+ {"dbsize",dbsizeCommand,1,REDIS_CMD_INLINE,0,0,0},
+ {"auth",authCommand,2,REDIS_CMD_INLINE,0,0,0},
+ {"ping",pingCommand,1,REDIS_CMD_INLINE,0,0,0},
+ {"echo",echoCommand,2,REDIS_CMD_BULK,0,0,0},
+ {"save",saveCommand,1,REDIS_CMD_INLINE,0,0,0},
+ {"bgsave",bgsaveCommand,1,REDIS_CMD_INLINE,0,0,0},
+ {"bgrewriteaof",bgrewriteaofCommand,1,REDIS_CMD_INLINE,0,0,0},
+ {"shutdown",shutdownCommand,1,REDIS_CMD_INLINE,0,0,0},
+ {"lastsave",lastsaveCommand,1,REDIS_CMD_INLINE,0,0,0},
+ {"type",typeCommand,2,REDIS_CMD_INLINE,1,1,1},
+ {"multi",multiCommand,1,REDIS_CMD_INLINE,0,0,0},
+ {"exec",execCommand,1,REDIS_CMD_INLINE,0,0,0},
+ {"discard",discardCommand,1,REDIS_CMD_INLINE,0,0,0},
+ {"sync",syncCommand,1,REDIS_CMD_INLINE,0,0,0},
+ {"flushdb",flushdbCommand,1,REDIS_CMD_INLINE,0,0,0},
+ {"flushall",flushallCommand,1,REDIS_CMD_INLINE,0,0,0},
+ {"sort",sortCommand,-2,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM,1,1,1},
+ {"info",infoCommand,1,REDIS_CMD_INLINE,0,0,0},
+ {"monitor",monitorCommand,1,REDIS_CMD_INLINE,0,0,0},
+ {"ttl",ttlCommand,2,REDIS_CMD_INLINE,1,1,1},
+ {"slaveof",slaveofCommand,3,REDIS_CMD_INLINE,0,0,0},
+ {"debug",debugCommand,-2,REDIS_CMD_INLINE,0,0,0},
+ {NULL,NULL,0,0,0,0,0}
};
/*============================ Utility functions ============================ */
};
/* Db->dict */
-static dictType hashDictType = {
+static dictType dbDictType = {
dictObjHash, /* hash function */
NULL, /* key dup */
NULL, /* val dup */
NULL /* val destructor */
};
+/* Hash type hash table (note that small hashes are represented with zimpaps) */
+static dictType hashDictType = {
+ dictEncObjHash, /* hash function */
+ NULL, /* key dup */
+ NULL, /* val dup */
+ dictEncObjKeyCompare, /* key compare */
+ dictRedisObjectDestructor, /* key destructor */
+ dictRedisObjectDestructor /* val destructor */
+};
+
/* Keylist hash table type has unencoded redis objects as keys and
* lists as values. It's used for blocking operations (BLPOP) and to
* map swapped keys to a list of clients waiting for this keys to be loaded. */
server.vm_max_memory = 1024LL*1024*1024*1; /* 1 GB of RAM */
server.vm_max_threads = 4;
server.vm_blocked_clients = 0;
+ server.hash_max_zipmap_entries = REDIS_HASH_MAX_ZIPMAP_ENTRIES;
+ server.hash_max_zipmap_value = REDIS_HASH_MAX_ZIPMAP_VALUE;
resetServerSaveParams();
exit(1);
}
for (j = 0; j < server.dbnum; j++) {
- server.db[j].dict = dictCreate(&hashDictType,NULL);
+ server.db[j].dict = dictCreate(&dbDictType,NULL);
server.db[j].expires = dictCreate(&keyptrDictType,NULL);
server.db[j].blockingkeys = dictCreate(&keylistDictType,NULL);
if (server.vm_enabled)
server.vm_pages = strtoll(argv[1], NULL, 10);
} else if (!strcasecmp(argv[0],"vm-max-threads") && argc == 2) {
server.vm_max_threads = strtoll(argv[1], NULL, 10);
+ } else if (!strcasecmp(argv[0],"hash-max-zipmap-entries") && argc == 2){
+ server.hash_max_zipmap_entries = strtol(argv[1], NULL, 10);
+ } else if (!strcasecmp(argv[0],"hash-max-zipmap-value") && argc == 2){
+ server.hash_max_zipmap_value = strtol(argv[1], NULL, 10);
+ } else if (!strcasecmp(argv[0],"vm-max-threads") && argc == 2) {
+ server.vm_max_threads = strtoll(argv[1], NULL, 10);
} else {
err = "Bad directive or wrong number of arguments"; goto loaderr;
}
}
/* Exec the command */
- if (c->flags & REDIS_MULTI && cmd->proc != execCommand) {
+ if (c->flags & REDIS_MULTI && cmd->proc != execCommand && cmd->proc != discardCommand) {
queueMultiCommand(c,cmd);
addReply(c,shared.queued);
} else {
} else {
return;
}
- processInputBuffer(c);
+ if (!(c->flags & REDIS_BLOCKED))
+ processInputBuffer(c);
}
static int selectDb(redisClient *c, int id) {
static void *dupClientReplyValue(void *o) {
incrRefCount((robj*)o);
- return 0;
+ return o;
}
static redisClient *createClient(int fd) {
(unsigned long) strlen(buf),buf));
}
+static void addReplyLong(redisClient *c, long l) {
+ char buf[128];
+ size_t len;
+
+ len = snprintf(buf,sizeof(buf),":%ld\r\n",l);
+ addReplySds(c,sdsnewlen(buf,len));
+}
+
static void addReplyBulkLen(redisClient *c, robj *obj) {
size_t len;
return createObject(REDIS_SET,d);
}
+static robj *createHashObject(void) {
+ /* All the Hashes start as zipmaps. Will be automatically converted
+ * into hash tables if there are enough elements or big elements
+ * inside. */
+ unsigned char *zm = zipmapNew();
+ robj *o = createObject(REDIS_HASH,zm);
+ o->encoding = REDIS_ENCODING_ZIPMAP;
+ return o;
+}
+
static robj *createZsetObject(void) {
zset *zs = zmalloc(sizeof(*zs));
}
static void freeHashObject(robj *o) {
- dictRelease((dict*) o->ptr);
+ switch (o->encoding) {
+ case REDIS_ENCODING_HT:
+ dictRelease((dict*) o->ptr);
+ break;
+ case REDIS_ENCODING_ZIPMAP:
+ zfree(o->ptr);
+ break;
+ default:
+ redisAssert(0);
+ break;
+ }
}
static void incrRefCount(robj *o) {
/* String objects in the form "2391" "-100" without any space and with a
* range of values that can fit in an 8, 16 or 32 bit signed value can be
* encoded as integers to save space */
-static int rdbTryIntegerEncoding(sds s, unsigned char *enc) {
+static int rdbTryIntegerEncoding(char *s, size_t len, unsigned char *enc) {
long long value;
char *endptr, buf[32];
/* If the number converted back into a string is not identical
* then it's not possible to encode the string as integer */
- if (strlen(buf) != sdslen(s) || memcmp(buf,s,sdslen(s))) return 0;
+ if (strlen(buf) != len || memcmp(buf,s,len)) return 0;
/* Finally check if it fits in our ranges */
if (value >= -(1<<7) && value <= (1<<7)-1) {
}
}
-static int rdbSaveLzfStringObject(FILE *fp, robj *obj) {
- unsigned int comprlen, outlen;
+static int rdbSaveLzfStringObject(FILE *fp, unsigned char *s, size_t len) {
+ size_t comprlen, outlen;
unsigned char byte;
void *out;
/* We require at least four bytes compression for this to be worth it */
- outlen = sdslen(obj->ptr)-4;
- if (outlen <= 0) return 0;
+ if (len <= 4) return 0;
+ outlen = len-4;
if ((out = zmalloc(outlen+1)) == NULL) return 0;
- comprlen = lzf_compress(obj->ptr, sdslen(obj->ptr), out, outlen);
+ comprlen = lzf_compress(s, len, out, outlen);
if (comprlen == 0) {
zfree(out);
return 0;
byte = (REDIS_RDB_ENCVAL<<6)|REDIS_RDB_ENC_LZF;
if (fwrite(&byte,1,1,fp) == 0) goto writeerr;
if (rdbSaveLen(fp,comprlen) == -1) goto writeerr;
- if (rdbSaveLen(fp,sdslen(obj->ptr)) == -1) goto writeerr;
+ if (rdbSaveLen(fp,len) == -1) goto writeerr;
if (fwrite(out,comprlen,1,fp) == 0) goto writeerr;
zfree(out);
return comprlen;
/* Save a string objet as [len][data] on disk. If the object is a string
* representation of an integer value we try to safe it in a special form */
-static int rdbSaveStringObjectRaw(FILE *fp, robj *obj) {
- size_t len;
+static int rdbSaveRawString(FILE *fp, unsigned char *s, size_t len) {
int enclen;
- len = sdslen(obj->ptr);
-
/* Try integer encoding */
if (len <= 11) {
unsigned char buf[5];
- if ((enclen = rdbTryIntegerEncoding(obj->ptr,buf)) > 0) {
+ if ((enclen = rdbTryIntegerEncoding((char*)s,len,buf)) > 0) {
if (fwrite(buf,enclen,1,fp) == 0) return -1;
return 0;
}
if (server.rdbcompression && len > 20) {
int retval;
- retval = rdbSaveLzfStringObject(fp,obj);
+ retval = rdbSaveLzfStringObject(fp,s,len);
if (retval == -1) return -1;
if (retval > 0) return 0;
/* retval == 0 means data can't be compressed, save the old way */
/* Store verbatim */
if (rdbSaveLen(fp,len) == -1) return -1;
- if (len && fwrite(obj->ptr,len,1,fp) == 0) return -1;
+ if (len && fwrite(s,len,1,fp) == 0) return -1;
return 0;
}
* this in order to avoid bugs) */
if (obj->encoding != REDIS_ENCODING_RAW) {
obj = getDecodedObject(obj);
- retval = rdbSaveStringObjectRaw(fp,obj);
+ retval = rdbSaveRawString(fp,obj->ptr,sdslen(obj->ptr));
decrRefCount(obj);
} else {
- retval = rdbSaveStringObjectRaw(fp,obj);
+ retval = rdbSaveRawString(fp,obj->ptr,sdslen(obj->ptr));
}
return retval;
}
if (rdbSaveDoubleValue(fp,*score) == -1) return -1;
}
dictReleaseIterator(di);
+ } else if (o->type == REDIS_HASH) {
+ /* Save a hash value */
+ if (o->encoding == REDIS_ENCODING_ZIPMAP) {
+ unsigned char *p = zipmapRewind(o->ptr);
+ unsigned int count = zipmapLen(o->ptr);
+ unsigned char *key, *val;
+ unsigned int klen, vlen;
+
+ if (rdbSaveLen(fp,count) == -1) return -1;
+ while((p = zipmapNext(p,&key,&klen,&val,&vlen)) != NULL) {
+ if (rdbSaveRawString(fp,key,klen) == -1) return -1;
+ if (rdbSaveRawString(fp,val,vlen) == -1) return -1;
+ }
+ } else {
+ dictIterator *di = dictGetIterator(o->ptr);
+ dictEntry *de;
+
+ if (rdbSaveLen(fp,dictSize((dict*)o->ptr)) == -1) return -1;
+ while((de = dictNext(di)) != NULL) {
+ robj *key = dictGetEntryKey(de);
+ robj *val = dictGetEntryVal(de);
+
+ if (rdbSaveStringObject(fp,key) == -1) return -1;
+ if (rdbSaveStringObject(fp,val) == -1) return -1;
+ }
+ dictReleaseIterator(di);
+ }
} else {
redisAssert(0 != 0);
}
static robj *rdbLoadObject(int type, FILE *fp) {
robj *o;
+ redisLog(REDIS_DEBUG,"LOADING OBJECT %d (at %d)\n",type,ftell(fp));
if (type == REDIS_STRING) {
/* Read string value */
if ((o = rdbLoadStringObject(fp)) == NULL) return NULL;
}
} else if (type == REDIS_ZSET) {
/* Read list/set value */
- uint32_t zsetlen;
+ size_t zsetlen;
zset *zs;
if ((zsetlen = rdbLoadLen(fp,NULL)) == REDIS_RDB_LENERR) return NULL;
zslInsert(zs->zsl,*score,ele);
incrRefCount(ele); /* added to skiplist */
}
+ } else if (type == REDIS_HASH) {
+ size_t hashlen;
+
+ if ((hashlen = rdbLoadLen(fp,NULL)) == REDIS_RDB_LENERR) return NULL;
+ o = createHashObject();
+ /* Too many entries? Use an hash table. */
+ if (hashlen > server.hash_max_zipmap_entries)
+ convertToRealHash(o);
+ /* Load every key/value, then set it into the zipmap or hash
+ * table, as needed. */
+ while(hashlen--) {
+ robj *key, *val;
+
+ if ((key = rdbLoadStringObject(fp)) == NULL) return NULL;
+ if ((val = rdbLoadStringObject(fp)) == NULL) return NULL;
+ /* If we are using a zipmap and there are too big values
+ * the object is converted to real hash table encoding. */
+ if (o->encoding != REDIS_ENCODING_HT &&
+ (sdslen(key->ptr) > server.hash_max_zipmap_value ||
+ sdslen(val->ptr) > server.hash_max_zipmap_value))
+ {
+ convertToRealHash(o);
+ }
+
+ if (o->encoding == REDIS_ENCODING_ZIPMAP) {
+ unsigned char *zm = o->ptr;
+
+ zm = zipmapSet(zm,key->ptr,sdslen(key->ptr),
+ val->ptr,sdslen(val->ptr),NULL);
+ o->ptr = zm;
+ decrRefCount(key);
+ decrRefCount(val);
+ } else {
+ tryObjectEncoding(key);
+ tryObjectEncoding(val);
+ dictAdd((dict*)o->ptr,key,val);
+ incrRefCount(key);
+ incrRefCount(val);
+ }
+ }
} else {
redisAssert(0 != 0);
}
* to overwrite the old. So we delete the old key in the database.
* This will also make sure that swap pages about the old object
* will be marked as free. */
- if (deleteIfSwapped(c->db,c->argv[1]))
+ if (server.vm_enabled && deleteIfSwapped(c->db,c->argv[1]))
incrRefCount(c->argv[1]);
dictReplace(c->db->dict,c->argv[1],c->argv[2]);
incrRefCount(c->argv[2]);
addReplySds(c,sdscatprintf(sdsempty(),":%lu\r\n",(unsigned long)totlen));
}
+static void substrCommand(redisClient *c) {
+ robj *o;
+ long start = atoi(c->argv[2]->ptr);
+ long end = atoi(c->argv[3]->ptr);
+
+ o = lookupKeyRead(c->db,c->argv[1]);
+ if (o == NULL) {
+ addReply(c,shared.nullbulk);
+ } else {
+ if (o->type != REDIS_STRING) {
+ addReply(c,shared.wrongtypeerr);
+ } else {
+ size_t rangelen, strlen;
+ sds range;
+
+ o = getDecodedObject(o);
+ strlen = sdslen(o->ptr);
+
+ /* convert negative indexes */
+ if (start < 0) start = strlen+start;
+ if (end < 0) end = strlen+end;
+ if (start < 0) start = 0;
+ if (end < 0) end = 0;
+
+ /* indexes sanity checks */
+ if (start > end || (size_t)start >= strlen) {
+ /* Out of range start or start > end result in null reply */
+ addReply(c,shared.nullbulk);
+ decrRefCount(o);
+ return;
+ }
+ if ((size_t)end >= strlen) end = strlen-1;
+ rangelen = (end-start)+1;
+
+ /* Return the result */
+ addReplySds(c,sdscatprintf(sdsempty(),"$%zu\r\n",rangelen));
+ range = sdsnewlen((char*)o->ptr+start,rangelen);
+ addReplySds(c,range);
+ addReply(c,shared.crlf);
+ decrRefCount(o);
+ }
+ }
+}
+
/* ========================= Type agnostic commands ========================= */
static void delCommand(redisClient *c) {
dictEntry *de;
sds pattern = c->argv[1]->ptr;
int plen = sdslen(pattern);
- unsigned long numkeys = 0, keyslen = 0;
+ unsigned long numkeys = 0;
robj *lenobj = createObject(REDIS_STRING,NULL);
di = dictGetIterator(c->db->dict);
if ((pattern[0] == '*' && pattern[1] == '\0') ||
stringmatchlen(pattern,plen,key,sdslen(key),0)) {
if (expireIfNeeded(c->db,keyobj) == 0) {
- if (numkeys != 0)
- addReply(c,shared.space);
+ addReplyBulkLen(c,keyobj);
addReply(c,keyobj);
+ addReply(c,shared.crlf);
numkeys++;
- keyslen += sdslen(key);
}
}
}
dictReleaseIterator(di);
- lenobj->ptr = sdscatprintf(sdsempty(),"$%lu\r\n",keyslen+(numkeys ? (numkeys-1) : 0));
- addReply(c,shared.crlf);
+ lenobj->ptr = sdscatprintf(sdsempty(),"*%lu\r\n",numkeys);
}
static void dbsizeCommand(redisClient *c) {
case REDIS_LIST: type = "+list"; break;
case REDIS_SET: type = "+set"; break;
case REDIS_ZSET: type = "+zset"; break;
- default: type = "unknown"; break;
+ case REDIS_HASH: type = "+hash"; break;
+ default: type = "+unknown"; break;
}
}
addReplySds(c,sdsnew(type));
lobj = lookupKeyWrite(c->db,c->argv[1]);
if (lobj == NULL) {
if (handleClientsWaitingListPush(c,c->argv[1],c->argv[2])) {
- addReply(c,shared.ok);
+ addReply(c,shared.cone);
return;
}
lobj = createListObject();
return;
}
if (handleClientsWaitingListPush(c,c->argv[1],c->argv[2])) {
- addReply(c,shared.ok);
+ addReply(c,shared.cone);
return;
}
list = lobj->ptr;
incrRefCount(c->argv[2]);
}
server.dirty++;
- addReply(c,shared.ok);
+ addReplySds(c,sdscatprintf(sdsempty(),":%d\r\n",listLength(list)));
}
static void lpushCommand(redisClient *c) {
#define REDIS_OP_UNION 0
#define REDIS_OP_DIFF 1
+#define REDIS_OP_INTER 2
static void sunionDiffGenericCommand(redisClient *c, robj **setskeys, int setsnum, robj *dstkey, int op) {
dict **dv = zmalloc(sizeof(dict*)*setsnum);
zskiplistNode *zn = zmalloc(sizeof(*zn));
zn->forward = zmalloc(sizeof(zskiplistNode*) * level);
+ if (level > 0)
+ zn->span = zmalloc(sizeof(unsigned int) * (level - 1));
zn->score = score;
zn->obj = obj;
return zn;
zsl->level = 1;
zsl->length = 0;
zsl->header = zslCreateNode(ZSKIPLIST_MAXLEVEL,0,NULL);
- for (j = 0; j < ZSKIPLIST_MAXLEVEL; j++)
+ for (j = 0; j < ZSKIPLIST_MAXLEVEL; j++) {
zsl->header->forward[j] = NULL;
+
+ /* span has space for ZSKIPLIST_MAXLEVEL-1 elements */
+ if (j < ZSKIPLIST_MAXLEVEL-1)
+ zsl->header->span[j] = 0;
+ }
zsl->header->backward = NULL;
zsl->tail = NULL;
return zsl;
static void zslFreeNode(zskiplistNode *node) {
decrRefCount(node->obj);
zfree(node->forward);
+ zfree(node->span);
zfree(node);
}
zskiplistNode *node = zsl->header->forward[0], *next;
zfree(zsl->header->forward);
+ zfree(zsl->header->span);
zfree(zsl->header);
while(node) {
next = node->forward[0];
static void zslInsert(zskiplist *zsl, double score, robj *obj) {
zskiplistNode *update[ZSKIPLIST_MAXLEVEL], *x;
+ unsigned int rank[ZSKIPLIST_MAXLEVEL];
int i, level;
x = zsl->header;
for (i = zsl->level-1; i >= 0; i--) {
+ /* store rank that is crossed to reach the insert position */
+ rank[i] = i == (zsl->level-1) ? 0 : rank[i+1];
+
while (x->forward[i] &&
(x->forward[i]->score < score ||
(x->forward[i]->score == score &&
- compareStringObjects(x->forward[i]->obj,obj) < 0)))
+ compareStringObjects(x->forward[i]->obj,obj) < 0))) {
+ rank[i] += i > 0 ? x->span[i-1] : 1;
x = x->forward[i];
+ }
update[i] = x;
}
/* we assume the key is not already inside, since we allow duplicated
* if the element is already inside or not. */
level = zslRandomLevel();
if (level > zsl->level) {
- for (i = zsl->level; i < level; i++)
+ for (i = zsl->level; i < level; i++) {
+ rank[i] = 0;
update[i] = zsl->header;
+ update[i]->span[i-1] = zsl->length;
+ }
zsl->level = level;
}
x = zslCreateNode(level,score,obj);
for (i = 0; i < level; i++) {
x->forward[i] = update[i]->forward[i];
update[i]->forward[i] = x;
+
+ /* update span covered by update[i] as x is inserted here */
+ if (i > 0) {
+ x->span[i-1] = update[i]->span[i-1] - (rank[0] - rank[i]);
+ update[i]->span[i-1] = (rank[0] - rank[i]) + 1;
+ }
}
+
+ /* increment span for untouched levels */
+ for (i = level; i < zsl->level; i++) {
+ update[i]->span[i-1]++;
+ }
+
x->backward = (update[0] == zsl->header) ? NULL : update[0];
if (x->forward[0])
x->forward[0]->backward = x;
zsl->length++;
}
+/* Internal function used by zslDelete, zslDeleteByScore and zslDeleteByRank */
+void zslDeleteNode(zskiplist *zsl, zskiplistNode *x, zskiplistNode **update) {
+ int i;
+ for (i = 0; i < zsl->level; i++) {
+ if (update[i]->forward[i] == x) {
+ if (i > 0) {
+ update[i]->span[i-1] += x->span[i-1] - 1;
+ }
+ update[i]->forward[i] = x->forward[i];
+ } else {
+ /* invariant: i > 0, because update[0]->forward[0]
+ * is always equal to x */
+ update[i]->span[i-1] -= 1;
+ }
+ }
+ if (x->forward[0]) {
+ x->forward[0]->backward = x->backward;
+ } else {
+ zsl->tail = x->backward;
+ }
+ while(zsl->level > 1 && zsl->header->forward[zsl->level-1] == NULL)
+ zsl->level--;
+ zsl->length--;
+}
+
/* Delete an element with matching score/object from the skiplist. */
static int zslDelete(zskiplist *zsl, double score, robj *obj) {
zskiplistNode *update[ZSKIPLIST_MAXLEVEL], *x;
* is to find the element with both the right score and object. */
x = x->forward[0];
if (x && score == x->score && compareStringObjects(x->obj,obj) == 0) {
- for (i = 0; i < zsl->level; i++) {
- if (update[i]->forward[i] != x) break;
- update[i]->forward[i] = x->forward[i];
- }
- if (x->forward[0]) {
- x->forward[0]->backward = (x->backward == zsl->header) ?
- NULL : x->backward;
- } else {
- zsl->tail = x->backward;
- }
+ zslDeleteNode(zsl, x, update);
zslFreeNode(x);
- while(zsl->level > 1 && zsl->header->forward[zsl->level-1] == NULL)
- zsl->level--;
- zsl->length--;
return 1;
} else {
return 0; /* not found */
* Min and mx are inclusive, so a score >= min || score <= max is deleted.
* Note that this function takes the reference to the hash table view of the
* sorted set, in order to remove the elements from the hash table too. */
-static unsigned long zslDeleteRange(zskiplist *zsl, double min, double max, dict *dict) {
+static unsigned long zslDeleteRangeByScore(zskiplist *zsl, double min, double max, dict *dict) {
zskiplistNode *update[ZSKIPLIST_MAXLEVEL], *x;
unsigned long removed = 0;
int i;
* is to find the element with both the right score and object. */
x = x->forward[0];
while (x && x->score <= max) {
- zskiplistNode *next;
+ zskiplistNode *next = x->forward[0];
+ zslDeleteNode(zsl, x, update);
+ dictDelete(dict,x->obj);
+ zslFreeNode(x);
+ removed++;
+ x = next;
+ }
+ return removed; /* not found */
+}
- for (i = 0; i < zsl->level; i++) {
- if (update[i]->forward[i] != x) break;
- update[i]->forward[i] = x->forward[i];
- }
- if (x->forward[0]) {
- x->forward[0]->backward = (x->backward == zsl->header) ?
- NULL : x->backward;
- } else {
- zsl->tail = x->backward;
+/* Delete all the elements with rank between start and end from the skiplist.
+ * Start and end are inclusive. Note that start and end need to be 1-based */
+static unsigned long zslDeleteRangeByRank(zskiplist *zsl, unsigned int start, unsigned int end, dict *dict) {
+ zskiplistNode *update[ZSKIPLIST_MAXLEVEL], *x;
+ unsigned long traversed = 0, removed = 0;
+ int i;
+
+ x = zsl->header;
+ for (i = zsl->level-1; i >= 0; i--) {
+ while (x->forward[i] && (traversed + (i > 0 ? x->span[i-1] : 1)) < start) {
+ traversed += i > 0 ? x->span[i-1] : 1;
+ x = x->forward[i];
}
- next = x->forward[0];
+ update[i] = x;
+ }
+
+ traversed++;
+ x = x->forward[0];
+ while (x && traversed <= end) {
+ zskiplistNode *next = x->forward[0];
+ zslDeleteNode(zsl, x, update);
dictDelete(dict,x->obj);
zslFreeNode(x);
- while(zsl->level > 1 && zsl->header->forward[zsl->level-1] == NULL)
- zsl->level--;
- zsl->length--;
removed++;
+ traversed++;
x = next;
}
- return removed; /* not found */
+ return removed;
}
/* Find the first node having a score equal or greater than the specified one.
return x->forward[0];
}
+/* Find the rank for an element by both score and key.
+ * Returns 0 when the element cannot be found, rank otherwise.
+ * Note that the rank is 1-based due to the span of zsl->header to the
+ * first element. */
+static unsigned long zslGetRank(zskiplist *zsl, double score, robj *o) {
+ zskiplistNode *x;
+ unsigned long rank = 0;
+ int i;
+
+ x = zsl->header;
+ for (i = zsl->level-1; i >= 0; i--) {
+ while (x->forward[i] &&
+ (x->forward[i]->score < score ||
+ (x->forward[i]->score == score &&
+ compareStringObjects(x->forward[i]->obj,o) <= 0))) {
+ rank += i > 0 ? x->span[i-1] : 1;
+ x = x->forward[i];
+ }
+
+ /* x might be equal to zsl->header, so test if obj is non-NULL */
+ if (x->obj && compareStringObjects(x->obj,o) == 0) {
+ return rank;
+ }
+ }
+ return 0;
+}
+
+/* Finds an element by its rank. The rank argument needs to be 1-based. */
+zskiplistNode* zslGetElementByRank(zskiplist *zsl, unsigned long rank) {
+ zskiplistNode *x;
+ unsigned long traversed = 0;
+ int i;
+
+ x = zsl->header;
+ for (i = zsl->level-1; i >= 0; i--) {
+ while (x->forward[i] && (traversed + (i > 0 ? x->span[i-1] : 1)) <= rank) {
+ traversed += i > 0 ? x->span[i-1] : 1;
+ x = x->forward[i];
+ }
+
+ if (traversed == rank) {
+ return x;
+ }
+ }
+ return NULL;
+}
+
/* The actual Z-commands implementations */
/* This generic command implements both ZADD and ZINCRBY.
return;
}
zs = zsetobj->ptr;
- deleted = zslDeleteRange(zs->zsl,min,max,zs->dict);
+ deleted = zslDeleteRangeByScore(zs->zsl,min,max,zs->dict);
if (htNeedsResize(zs->dict)) dictResize(zs->dict);
server.dirty += deleted;
addReplySds(c,sdscatprintf(sdsempty(),":%lu\r\n",deleted));
}
}
+static void zremrangebyrankCommand(redisClient *c) {
+ int start = atoi(c->argv[2]->ptr);
+ int end = atoi(c->argv[3]->ptr);
+ robj *zsetobj;
+ zset *zs;
+
+ zsetobj = lookupKeyWrite(c->db,c->argv[1]);
+ if (zsetobj == NULL) {
+ addReply(c,shared.czero);
+ } else {
+ if (zsetobj->type != REDIS_ZSET) {
+ addReply(c,shared.wrongtypeerr);
+ return;
+ }
+
+ zs = zsetobj->ptr;
+ int llen = zs->zsl->length;
+ long deleted;
+
+ /* convert negative indexes */
+ if (start < 0) start = llen+start;
+ if (end < 0) end = llen+end;
+ if (start < 0) start = 0;
+ if (end < 0) end = 0;
+
+ /* indexes sanity checks */
+ if (start > end || start >= llen) {
+ addReply(c,shared.czero);
+ return;
+ }
+ if (end >= llen) end = llen-1;
+
+ /* increment start and end because zsl*Rank functions
+ * use 1-based rank */
+ deleted = zslDeleteRangeByRank(zs->zsl,start+1,end+1,zs->dict);
+ if (htNeedsResize(zs->dict)) dictResize(zs->dict);
+ server.dirty += deleted;
+ addReplyLong(c, deleted);
+ }
+}
+
+typedef struct {
+ dict *dict;
+ double weight;
+} zsetopsrc;
+
+static int qsortCompareZsetopsrcByCardinality(const void *s1, const void *s2) {
+ zsetopsrc *d1 = (void*) s1, *d2 = (void*) s2;
+ unsigned long size1, size2;
+ size1 = d1->dict ? dictSize(d1->dict) : 0;
+ size2 = d2->dict ? dictSize(d2->dict) : 0;
+ return size1 - size2;
+}
+
+static void zunionInterGenericCommand(redisClient *c, robj *dstkey, int op) {
+ int i, j, zsetnum;
+ zsetopsrc *src;
+ robj *dstobj;
+ zset *dstzset;
+ dictIterator *di;
+ dictEntry *de;
+
+ /* expect zsetnum input keys to be given */
+ zsetnum = atoi(c->argv[2]->ptr);
+ if (zsetnum < 1) {
+ addReplySds(c,sdsnew("-ERR at least 1 input key is needed for ZUNION/ZINTER\r\n"));
+ return;
+ }
+
+ /* test if the expected number of keys would overflow */
+ if (3+zsetnum > c->argc) {
+ addReply(c,shared.syntaxerr);
+ return;
+ }
+
+ /* read keys to be used for input */
+ src = zmalloc(sizeof(zsetopsrc) * zsetnum);
+ for (i = 0, j = 3; i < zsetnum; i++, j++) {
+ robj *zsetobj = lookupKeyWrite(c->db,c->argv[j]);
+ if (!zsetobj) {
+ src[i].dict = NULL;
+ } else {
+ if (zsetobj->type != REDIS_ZSET) {
+ zfree(src);
+ addReply(c,shared.wrongtypeerr);
+ return;
+ }
+ src[i].dict = ((zset*)zsetobj->ptr)->dict;
+ }
+
+ /* default all weights to 1 */
+ src[i].weight = 1.0;
+ }
+
+ /* parse optional extra arguments */
+ if (j < c->argc) {
+ int remaining = c->argc-j;
+
+ while (remaining) {
+ if (!strcasecmp(c->argv[j]->ptr,"weights")) {
+ j++; remaining--;
+ if (remaining < zsetnum) {
+ zfree(src);
+ addReplySds(c,sdsnew("-ERR not enough weights for ZUNION/ZINTER\r\n"));
+ return;
+ }
+ for (i = 0; i < zsetnum; i++, j++, remaining--) {
+ src[i].weight = strtod(c->argv[j]->ptr, NULL);
+ }
+ } else {
+ zfree(src);
+ addReply(c,shared.syntaxerr);
+ return;
+ }
+ }
+ }
+
+ dstobj = createZsetObject();
+ dstzset = dstobj->ptr;
+
+ if (op == REDIS_OP_INTER) {
+ /* sort sets from the smallest to largest, this will improve our
+ * algorithm's performance */
+ qsort(src,zsetnum,sizeof(zsetopsrc), qsortCompareZsetopsrcByCardinality);
+
+ /* skip going over all entries if the smallest zset is NULL or empty */
+ if (src[0].dict && dictSize(src[0].dict) > 0) {
+ /* precondition: as src[0].dict is non-empty and the zsets are ordered
+ * from small to large, all src[i > 0].dict are non-empty too */
+ di = dictGetIterator(src[0].dict);
+ while((de = dictNext(di)) != NULL) {
+ double *score = zmalloc(sizeof(double));
+ *score = 0.0;
+
+ for (j = 0; j < zsetnum; j++) {
+ dictEntry *other = (j == 0) ? de : dictFind(src[j].dict,dictGetEntryKey(de));
+ if (other) {
+ *score = *score + src[j].weight * (*(double*)dictGetEntryVal(other));
+ } else {
+ break;
+ }
+ }
+
+ /* skip entry when not present in every source dict */
+ if (j != zsetnum) {
+ zfree(score);
+ } else {
+ robj *o = dictGetEntryKey(de);
+ dictAdd(dstzset->dict,o,score);
+ incrRefCount(o); /* added to dictionary */
+ zslInsert(dstzset->zsl,*score,o);
+ incrRefCount(o); /* added to skiplist */
+ }
+ }
+ dictReleaseIterator(di);
+ }
+ } else if (op == REDIS_OP_UNION) {
+ for (i = 0; i < zsetnum; i++) {
+ if (!src[i].dict) continue;
+
+ di = dictGetIterator(src[i].dict);
+ while((de = dictNext(di)) != NULL) {
+ /* skip key when already processed */
+ if (dictFind(dstzset->dict,dictGetEntryKey(de)) != NULL) continue;
+
+ double *score = zmalloc(sizeof(double));
+ *score = 0.0;
+ for (j = 0; j < zsetnum; j++) {
+ if (!src[j].dict) continue;
+
+ dictEntry *other = (i == j) ? de : dictFind(src[j].dict,dictGetEntryKey(de));
+ if (other) {
+ *score = *score + src[j].weight * (*(double*)dictGetEntryVal(other));
+ }
+ }
+
+ robj *o = dictGetEntryKey(de);
+ dictAdd(dstzset->dict,o,score);
+ incrRefCount(o); /* added to dictionary */
+ zslInsert(dstzset->zsl,*score,o);
+ incrRefCount(o); /* added to skiplist */
+ }
+ dictReleaseIterator(di);
+ }
+ } else {
+ /* unknown operator */
+ redisAssert(op == REDIS_OP_INTER || op == REDIS_OP_UNION);
+ }
+
+ deleteKey(c->db,dstkey);
+ dictAdd(c->db->dict,dstkey,dstobj);
+ incrRefCount(dstkey);
+
+ addReplyLong(c, dstzset->zsl->length);
+ server.dirty++;
+ zfree(src);
+}
+
+static void zunionCommand(redisClient *c) {
+ zunionInterGenericCommand(c,c->argv[1], REDIS_OP_UNION);
+}
+
+static void zinterCommand(redisClient *c) {
+ zunionInterGenericCommand(c,c->argv[1], REDIS_OP_INTER);
+}
+
static void zrangeGenericCommand(redisClient *c, int reverse) {
robj *o;
int start = atoi(c->argv[2]->ptr);
if (end >= llen) end = llen-1;
rangelen = (end-start)+1;
- /* Return the result in form of a multi-bulk reply */
+ /* check if starting point is trivial, before searching
+ * the element in log(N) time */
if (reverse) {
- ln = zsl->tail;
- while (start--)
- ln = ln->backward;
+ ln = start == 0 ? zsl->tail : zslGetElementByRank(zsl, llen-start);
} else {
- ln = zsl->header->forward[0];
- while (start--)
- ln = ln->forward[0];
+ ln = start == 0 ? zsl->header->forward[0] : zslGetElementByRank(zsl, start+1);
}
+ /* Return the result in form of a multi-bulk reply */
addReplySds(c,sdscatprintf(sdsempty(),"*%d\r\n",
withscores ? (rangelen*2) : rangelen));
for (j = 0; j < rangelen; j++) {
zrangeGenericCommand(c,1);
}
-static void zrangebyscoreCommand(redisClient *c) {
+/* This command implements both ZRANGEBYSCORE and ZCOUNT.
+ * If justcount is non-zero, just the count is returned. */
+static void genericZrangebyscoreCommand(redisClient *c, int justcount) {
robj *o;
- double min = strtod(c->argv[2]->ptr,NULL);
- double max = strtod(c->argv[3]->ptr,NULL);
+ double min, max;
+ int minex = 0, maxex = 0; /* are min or max exclusive? */
int offset = 0, limit = -1;
+ int withscores = 0;
+ int badsyntax = 0;
+
+ /* Parse the min-max interval. If one of the values is prefixed
+ * by the "(" character, it's considered "open". For instance
+ * ZRANGEBYSCORE zset (1.5 (2.5 will match min < x < max
+ * ZRANGEBYSCORE zset 1.5 2.5 will instead match min <= x <= max */
+ if (((char*)c->argv[2]->ptr)[0] == '(') {
+ min = strtod((char*)c->argv[2]->ptr+1,NULL);
+ minex = 1;
+ } else {
+ min = strtod(c->argv[2]->ptr,NULL);
+ }
+ if (((char*)c->argv[3]->ptr)[0] == '(') {
+ max = strtod((char*)c->argv[3]->ptr+1,NULL);
+ maxex = 1;
+ } else {
+ max = strtod(c->argv[3]->ptr,NULL);
+ }
- if (c->argc != 4 && c->argc != 7) {
+ /* Parse "WITHSCORES": note that if the command was called with
+ * the name ZCOUNT then we are sure that c->argc == 4, so we'll never
+ * enter the following paths to parse WITHSCORES and LIMIT. */
+ if (c->argc == 5 || c->argc == 8) {
+ if (strcasecmp(c->argv[c->argc-1]->ptr,"withscores") == 0)
+ withscores = 1;
+ else
+ badsyntax = 1;
+ }
+ if (c->argc != (4 + withscores) && c->argc != (7 + withscores))
+ badsyntax = 1;
+ if (badsyntax) {
addReplySds(c,
sdsnew("-ERR wrong number of arguments for ZRANGEBYSCORE\r\n"));
return;
- } else if (c->argc == 7 && strcasecmp(c->argv[4]->ptr,"limit")) {
+ }
+
+ /* Parse "LIMIT" */
+ if (c->argc == (7 + withscores) && strcasecmp(c->argv[4]->ptr,"limit")) {
addReply(c,shared.syntaxerr);
return;
- } else if (c->argc == 7) {
+ } else if (c->argc == (7 + withscores)) {
offset = atoi(c->argv[5]->ptr);
limit = atoi(c->argv[6]->ptr);
if (offset < 0) offset = 0;
}
+ /* Ok, lookup the key and get the range */
o = lookupKeyRead(c->db,c->argv[1]);
if (o == NULL) {
- addReply(c,shared.nullmultibulk);
+ addReply(c,justcount ? shared.czero : shared.nullmultibulk);
} else {
if (o->type != REDIS_ZSET) {
addReply(c,shared.wrongtypeerr);
zset *zsetobj = o->ptr;
zskiplist *zsl = zsetobj->zsl;
zskiplistNode *ln;
- robj *ele, *lenobj;
- unsigned int rangelen = 0;
+ robj *ele, *lenobj = NULL;
+ unsigned long rangelen = 0;
- /* Get the first node with the score >= min */
+ /* Get the first node with the score >= min, or with
+ * score > min if 'minex' is true. */
ln = zslFirstWithScore(zsl,min);
+ while (minex && ln && ln->score == min) ln = ln->forward[0];
+
if (ln == NULL) {
/* No element matching the speciifed interval */
- addReply(c,shared.emptymultibulk);
+ addReply(c,justcount ? shared.czero : shared.emptymultibulk);
return;
}
* are in the list, so we push this object that will represent
* the multi-bulk length in the output buffer, and will "fix"
* it later */
- lenobj = createObject(REDIS_STRING,NULL);
- addReply(c,lenobj);
- decrRefCount(lenobj);
+ if (!justcount) {
+ lenobj = createObject(REDIS_STRING,NULL);
+ addReply(c,lenobj);
+ decrRefCount(lenobj);
+ }
- while(ln && ln->score <= max) {
+ while(ln && (maxex ? (ln->score < max) : (ln->score <= max))) {
if (offset) {
offset--;
ln = ln->forward[0];
continue;
}
if (limit == 0) break;
- ele = ln->obj;
- addReplyBulkLen(c,ele);
- addReply(c,ele);
- addReply(c,shared.crlf);
+ if (!justcount) {
+ ele = ln->obj;
+ addReplyBulkLen(c,ele);
+ addReply(c,ele);
+ addReply(c,shared.crlf);
+ if (withscores)
+ addReplyDouble(c,ln->score);
+ }
ln = ln->forward[0];
rangelen++;
if (limit > 0) limit--;
}
- lenobj->ptr = sdscatprintf(sdsempty(),"*%d\r\n",rangelen);
+ if (justcount) {
+ addReplyLong(c,(long)rangelen);
+ } else {
+ lenobj->ptr = sdscatprintf(sdsempty(),"*%lu\r\n",
+ withscores ? (rangelen*2) : rangelen);
+ }
}
}
}
+static void zrangebyscoreCommand(redisClient *c) {
+ genericZrangebyscoreCommand(c,0);
+}
+
+static void zcountCommand(redisClient *c) {
+ genericZrangebyscoreCommand(c,1);
+}
+
static void zcardCommand(redisClient *c) {
robj *o;
zset *zs;
}
}
+static void zrankGenericCommand(redisClient *c, int reverse) {
+ robj *o;
+ o = lookupKeyRead(c->db,c->argv[1]);
+ if (o == NULL) {
+ addReply(c,shared.nullbulk);
+ return;
+ }
+ if (o->type != REDIS_ZSET) {
+ addReply(c,shared.wrongtypeerr);
+ } else {
+ zset *zs = o->ptr;
+ zskiplist *zsl = zs->zsl;
+ dictEntry *de;
+ unsigned long rank;
+
+ de = dictFind(zs->dict,c->argv[2]);
+ if (!de) {
+ addReply(c,shared.nullbulk);
+ return;
+ }
+
+ double *score = dictGetEntryVal(de);
+ rank = zslGetRank(zsl, *score, c->argv[2]);
+ if (rank) {
+ if (reverse) {
+ addReplyLong(c, zsl->length - rank);
+ } else {
+ addReplyLong(c, rank-1);
+ }
+ } else {
+ addReply(c,shared.nullbulk);
+ }
+ }
+}
+
+static void zrankCommand(redisClient *c) {
+ zrankGenericCommand(c, 0);
+}
+
+static void zrevrankCommand(redisClient *c) {
+ zrankGenericCommand(c, 1);
+}
+
+/* =================================== Hashes =============================== */
+static void hsetCommand(redisClient *c) {
+ int update = 0;
+ robj *o = lookupKeyWrite(c->db,c->argv[1]);
+
+ if (o == NULL) {
+ o = createHashObject();
+ dictAdd(c->db->dict,c->argv[1],o);
+ incrRefCount(c->argv[1]);
+ } else {
+ if (o->type != REDIS_HASH) {
+ addReply(c,shared.wrongtypeerr);
+ return;
+ }
+ }
+ /* We want to convert the zipmap into an hash table right now if the
+ * entry to be added is too big. Note that we check if the object
+ * is integer encoded before to try fetching the length in the test below.
+ * This is because integers are small, but currently stringObjectLen()
+ * performs a slow conversion: not worth it. */
+ if (o->encoding == REDIS_ENCODING_ZIPMAP &&
+ ((c->argv[2]->encoding == REDIS_ENCODING_RAW &&
+ sdslen(c->argv[2]->ptr) > server.hash_max_zipmap_value) ||
+ (c->argv[3]->encoding == REDIS_ENCODING_RAW &&
+ sdslen(c->argv[3]->ptr) > server.hash_max_zipmap_value)))
+ {
+ convertToRealHash(o);
+ }
+
+ if (o->encoding == REDIS_ENCODING_ZIPMAP) {
+ unsigned char *zm = o->ptr;
+ robj *valobj = getDecodedObject(c->argv[3]);
+
+ zm = zipmapSet(zm,c->argv[2]->ptr,sdslen(c->argv[2]->ptr),
+ valobj->ptr,sdslen(valobj->ptr),&update);
+ decrRefCount(valobj);
+ o->ptr = zm;
+
+ /* And here there is the second check for hash conversion...
+ * we want to do it only if the operation was not just an update as
+ * zipmapLen() is O(N). */
+ if (!update && zipmapLen(zm) > server.hash_max_zipmap_entries)
+ convertToRealHash(o);
+ } else {
+ tryObjectEncoding(c->argv[2]);
+ /* note that c->argv[3] is already encoded, as the latest arg
+ * of a bulk command is always integer encoded if possible. */
+ if (dictAdd(o->ptr,c->argv[2],c->argv[3]) == DICT_OK) {
+ incrRefCount(c->argv[2]);
+ } else {
+ update = 1;
+ }
+ incrRefCount(c->argv[3]);
+ }
+ server.dirty++;
+ addReplySds(c,sdscatprintf(sdsempty(),":%d\r\n",update == 0));
+}
+
+static void hgetCommand(redisClient *c) {
+ robj *o = lookupKeyRead(c->db,c->argv[1]);
+
+ if (o == NULL) {
+ addReply(c,shared.nullbulk);
+ return;
+ } else {
+ if (o->type != REDIS_HASH) {
+ addReply(c,shared.wrongtypeerr);
+ return;
+ }
+
+ if (o->encoding == REDIS_ENCODING_ZIPMAP) {
+ unsigned char *zm = o->ptr;
+ unsigned char *val;
+ unsigned int vlen;
+
+ if (zipmapGet(zm,c->argv[2]->ptr,sdslen(c->argv[2]->ptr), &val,&vlen)) {
+ addReplySds(c,sdscatprintf(sdsempty(),"$%u\r\n", vlen));
+ addReplySds(c,sdsnewlen(val,vlen));
+ addReply(c,shared.crlf);
+ return;
+ } else {
+ addReply(c,shared.nullbulk);
+ return;
+ }
+ } else {
+ struct dictEntry *de;
+
+ de = dictFind(o->ptr,c->argv[2]);
+ if (de == NULL) {
+ addReply(c,shared.nullbulk);
+ } else {
+ robj *e = dictGetEntryVal(de);
+
+ addReplyBulkLen(c,e);
+ addReply(c,e);
+ addReply(c,shared.crlf);
+ }
+ }
+ }
+}
+
+static void hdelCommand(redisClient *c) {
+ robj *o = lookupKeyRead(c->db,c->argv[1]);
+
+ if (o == NULL) {
+ addReply(c,shared.czero);
+ return;
+ } else {
+ int deleted = 0;
+
+ if (o->type != REDIS_HASH) {
+ addReply(c,shared.wrongtypeerr);
+ return;
+ }
+
+ if (o->encoding == REDIS_ENCODING_ZIPMAP) {
+ o->ptr = zipmapDel((unsigned char*) o->ptr,
+ (unsigned char*) c->argv[2]->ptr,
+ sdslen(c->argv[2]->ptr), &deleted);
+ } else {
+ deleted = dictDelete((dict*)o->ptr,c->argv[2]) == DICT_OK;
+ }
+ addReply(c,deleted ? shared.cone : shared.czero);
+ }
+}
+
+static void convertToRealHash(robj *o) {
+ unsigned char *key, *val, *p, *zm = o->ptr;
+ unsigned int klen, vlen;
+ dict *dict = dictCreate(&hashDictType,NULL);
+
+ assert(o->type == REDIS_HASH && o->encoding != REDIS_ENCODING_HT);
+ p = zipmapRewind(zm);
+ while((p = zipmapNext(p,&key,&klen,&val,&vlen)) != NULL) {
+ robj *keyobj, *valobj;
+
+ keyobj = createStringObject((char*)key,klen);
+ valobj = createStringObject((char*)val,vlen);
+ tryObjectEncoding(keyobj);
+ tryObjectEncoding(valobj);
+ dictAdd(dict,keyobj,valobj);
+ }
+ o->encoding = REDIS_ENCODING_HT;
+ o->ptr = dict;
+ zfree(zm);
+}
+
/* ========================= Non type-specific commands ==================== */
static void flushdbCommand(redisClient *c) {
time_t uptime = time(NULL)-server.stat_starttime;
int j;
char hmem[64];
+
+ server.hash_max_zipmap_entries = REDIS_HASH_MAX_ZIPMAP_ENTRIES;
+ server.hash_max_zipmap_value = REDIS_HASH_MAX_ZIPMAP_VALUE;
bytesToHuman(hmem,zmalloc_used_memory());
info = sdscatprintf(sdsempty(),
"bgrewriteaof_in_progress:%d\r\n"
"total_connections_received:%lld\r\n"
"total_commands_processed:%lld\r\n"
+ "hash_max_zipmap_entries:%ld\r\n"
+ "hash_max_zipmap_value:%ld\r\n"
"vm_enabled:%d\r\n"
"role:%s\r\n"
,REDIS_VERSION,
server.bgrewritechildpid != -1,
server.stat_numconnections,
server.stat_numcommands,
+ server.hash_max_zipmap_entries,
+ server.hash_max_zipmap_value,
server.vm_enabled != 0,
server.masterhost == NULL ? "master" : "slave"
);
addReply(c,shared.ok);
}
+static void discardCommand(redisClient *c) {
+ if (!(c->flags & REDIS_MULTI)) {
+ addReplySds(c,sdsnew("-ERR DISCARD without MULTI\r\n"));
+ return;
+ }
+
+ freeClientMultiState(c);
+ initClientMultiState(c);
+ c->flags &= (~REDIS_MULTI);
+ addReply(c,shared.ok);
+}
+
static void execCommand(redisClient *c) {
int j;
robj **orig_argv;
}
/* Mark the client as a blocked client */
c->flags |= REDIS_BLOCKED;
- aeDeleteFileEvent(server.el,c->fd,AE_READABLE);
server.blpop_blocked_clients++;
}
c->blockingkeys = NULL;
c->flags &= (~REDIS_BLOCKED);
server.blpop_blocked_clients--;
- /* Ok now we are ready to get read events from socket, note that we
- * can't trap errors here as it's possible that unblockClientWaitingDatas() is
- * called from freeClient() itself, and the only thing we can do
- * if we failed to register the READABLE event is to kill the client.
- * Still the following function should never fail in the real world as
- * we are sure the file descriptor is sane, and we exit on out of mem. */
- aeCreateFileEvent(server.el, c->fd, AE_READABLE, readQueryFromClient, c);
- /* As a final step we want to process data if there is some command waiting
+ /* We want to process data if there is some command waiting
* in the input buffer. Note that this is safe even if
* unblockClientWaitingData() gets called from freeClient() because
* freeClient() will be smart enough to call this function
static int syncWithMaster(void) {
char buf[1024], tmpfile[256], authcmd[1024];
- int dumpsize;
+ long dumpsize;
int fd = anetTcpConnect(NULL,server.masterhost,server.masterport);
- int dfd;
+ int dfd, maxtries = 5;
if (fd == -1) {
redisLog(REDIS_WARNING,"Unable to connect to MASTER: %s",
redisLog(REDIS_WARNING,"Bad protocol from MASTER, the first byte is not '$', are you sure the host and port are right?");
return REDIS_ERR;
}
- dumpsize = atoi(buf+1);
- redisLog(REDIS_NOTICE,"Receiving %d bytes data dump from MASTER",dumpsize);
+ dumpsize = strtol(buf+1,NULL,10);
+ redisLog(REDIS_NOTICE,"Receiving %ld bytes data dump from MASTER",dumpsize);
/* Read the bulk write data on a temp file */
- snprintf(tmpfile,256,"temp-%d.%ld.rdb",(int)time(NULL),(long int)random());
- dfd = open(tmpfile,O_CREAT|O_WRONLY,0644);
+ while(maxtries--) {
+ snprintf(tmpfile,256,
+ "temp-%d.%ld.rdb",(int)time(NULL),(long int)getpid());
+ dfd = open(tmpfile,O_CREAT|O_WRONLY|O_EXCL,0644);
+ if (dfd != -1) break;
+ sleep(1);
+ }
if (dfd == -1) {
close(fd);
redisLog(REDIS_WARNING,"Opening the temp file needed for MASTER <-> SLAVE synchronization: %s",strerror(errno));
}
/* Write an object into a file in the bulk format $<count>\r\n<payload>\r\n */
-static int fwriteBulk(FILE *fp, robj *obj) {
+static int fwriteBulkObject(FILE *fp, robj *obj) {
char buf[128];
int decrrc = 0;
return 0;
}
+/* Write binary-safe string into a file in the bulkformat
+ * $<count>\r\n<payload>\r\n */
+static int fwriteBulkString(FILE *fp, char *s, unsigned long len) {
+ char buf[128];
+
+ snprintf(buf,sizeof(buf),"$%ld\r\n",(unsigned long)len);
+ if (fwrite(buf,strlen(buf),1,fp) == 0) return 0;
+ if (len && fwrite(s,len,1,fp) == 0) return 0;
+ if (fwrite("\r\n",2,1,fp) == 0) return 0;
+ return 1;
+}
+
/* Write a double value in bulk format $<count>\r\n<payload>\r\n */
static int fwriteBulkDouble(FILE *fp, double d) {
char buf[128], dbuf[128];
char cmd[]="*3\r\n$3\r\nSET\r\n";
if (fwrite(cmd,sizeof(cmd)-1,1,fp) == 0) goto werr;
/* Key and value */
- if (fwriteBulk(fp,key) == 0) goto werr;
- if (fwriteBulk(fp,o) == 0) goto werr;
+ if (fwriteBulkObject(fp,key) == 0) goto werr;
+ if (fwriteBulkObject(fp,o) == 0) goto werr;
} else if (o->type == REDIS_LIST) {
/* Emit the RPUSHes needed to rebuild the list */
list *list = o->ptr;
robj *eleobj = listNodeValue(ln);
if (fwrite(cmd,sizeof(cmd)-1,1,fp) == 0) goto werr;
- if (fwriteBulk(fp,key) == 0) goto werr;
- if (fwriteBulk(fp,eleobj) == 0) goto werr;
+ if (fwriteBulkObject(fp,key) == 0) goto werr;
+ if (fwriteBulkObject(fp,eleobj) == 0) goto werr;
}
} else if (o->type == REDIS_SET) {
/* Emit the SADDs needed to rebuild the set */
robj *eleobj = dictGetEntryKey(de);
if (fwrite(cmd,sizeof(cmd)-1,1,fp) == 0) goto werr;
- if (fwriteBulk(fp,key) == 0) goto werr;
- if (fwriteBulk(fp,eleobj) == 0) goto werr;
+ if (fwriteBulkObject(fp,key) == 0) goto werr;
+ if (fwriteBulkObject(fp,eleobj) == 0) goto werr;
}
dictReleaseIterator(di);
} else if (o->type == REDIS_ZSET) {
double *score = dictGetEntryVal(de);
if (fwrite(cmd,sizeof(cmd)-1,1,fp) == 0) goto werr;
- if (fwriteBulk(fp,key) == 0) goto werr;
+ if (fwriteBulkObject(fp,key) == 0) goto werr;
if (fwriteBulkDouble(fp,*score) == 0) goto werr;
- if (fwriteBulk(fp,eleobj) == 0) goto werr;
+ if (fwriteBulkObject(fp,eleobj) == 0) goto werr;
}
dictReleaseIterator(di);
+ } else if (o->type == REDIS_HASH) {
+ char cmd[]="*4\r\n$4\r\nHSET\r\n";
+
+ /* Emit the HSETs needed to rebuild the hash */
+ if (o->encoding == REDIS_ENCODING_ZIPMAP) {
+ unsigned char *p = zipmapRewind(o->ptr);
+ unsigned char *field, *val;
+ unsigned int flen, vlen;
+
+ while((p = zipmapNext(p,&field,&flen,&val,&vlen)) != NULL) {
+ if (fwrite(cmd,sizeof(cmd)-1,1,fp) == 0) goto werr;
+ if (fwriteBulkObject(fp,key) == 0) goto werr;
+ if (fwriteBulkString(fp,(char*)field,flen) == -1)
+ return -1;
+ if (fwriteBulkString(fp,(char*)val,vlen) == -1)
+ return -1;
+ }
+ } else {
+ dictIterator *di = dictGetIterator(o->ptr);
+ dictEntry *de;
+
+ while((de = dictNext(di)) != NULL) {
+ robj *field = dictGetEntryKey(de);
+ robj *val = dictGetEntryVal(de);
+
+ if (fwrite(cmd,sizeof(cmd)-1,1,fp) == 0) goto werr;
+ if (fwriteBulkObject(fp,key) == 0) goto werr;
+ if (fwriteBulkObject(fp,field) == -1) return -1;
+ if (fwriteBulkObject(fp,val) == -1) return -1;
+ }
+ dictReleaseIterator(di);
+ }
} else {
redisAssert(0 != 0);
}
/* If this key is already expired skip it */
if (expiretime < now) continue;
if (fwrite(cmd,sizeof(cmd)-1,1,fp) == 0) goto werr;
- if (fwriteBulk(fp,key) == 0) goto werr;
+ if (fwriteBulkObject(fp,key) == 0) goto werr;
if (fwriteBulkLong(fp,expiretime) == 0) goto werr;
}
if (swapped) decrRefCount(o);
int bit = page&7;
redisAssert(vmFreePage(page) == 1);
server.vm_bitmap[byte] |= 1<<bit;
- redisLog(REDIS_DEBUG,"Mark used: %lld (byte:%lld bit:%d)\n",
- (long long)page, (long long)byte, bit);
}
/* Mark N contiguous pages as used, with 'page' being the first. */
for (j = 0; j < count; j++)
vmMarkPageUsed(page+j);
server.vm_stats_used_pages += count;
+ redisLog(REDIS_DEBUG,"Mark USED pages: %lld pages at %lld\n",
+ (long long)count, (long long)page);
}
/* Mark the page as free */
int bit = page&7;
redisAssert(vmFreePage(page) == 0);
server.vm_bitmap[byte] &= ~(1<<bit);
- redisLog(REDIS_DEBUG,"Mark free: %lld (byte:%lld bit:%d)\n",
- (long long)page, (long long)byte, bit);
}
/* Mark N contiguous pages as free, with 'page' being the first. */
for (j = 0; j < count; j++)
vmMarkPageFree(page+j);
server.vm_stats_used_pages -= count;
- if (server.vm_stats_used_pages > 100000000) {
- *((char*)-1) = 'x';
- }
+ redisLog(REDIS_DEBUG,"Mark FREE pages: %lld pages at %lld\n",
+ (long long)count, (long long)page);
}
/* Test if the page is free */
numfree = 0;
}
}
- redisLog(REDIS_DEBUG, "THIS: %lld (%c)\n", (long long) this, vmFreePage(this) ? 'F' : 'X');
if (vmFreePage(this)) {
/* This is a free page */
numfree++;
if (numfree == n) {
*first = this-(n-1);
server.vm_next_page = this+1;
+ redisLog(REDIS_DEBUG, "FOUND CONTIGUOUS PAGES: %lld pages at %lld\n", (long long) n, (long long) *first);
return REDIS_OK;
}
} else {
if (fseeko(server.vm_fp,page*server.vm_page_size,SEEK_SET) == -1) {
if (server.vm_enabled) pthread_mutex_unlock(&server.io_swapfile_mutex);
redisLog(REDIS_WARNING,
- "Critical VM problem in vmSwapObjectBlocking(): can't seek: %s",
+ "Critical VM problem in vmWriteObjectOnSwap(): can't seek: %s",
strerror(errno));
return REDIS_ERR;
}
rdbSaveObject(server.vm_fp,o);
+ fflush(server.vm_fp);
if (server.vm_enabled) pthread_mutex_unlock(&server.io_swapfile_mutex);
return REDIS_OK;
}
(unsigned long long) page, (unsigned long long) pages);
server.vm_stats_swapped_objects++;
server.vm_stats_swapouts++;
- fflush(server.vm_fp);
return REDIS_OK;
}
}
break;
}
- return (double)asize*log(1+asize);
+ return (double)age*log(1+asize);
}
/* Try to swap an object that's a good candidate for swapping.
}
}
}
- if (best == NULL) {
- redisLog(REDIS_DEBUG,"No swappable key found!");
- return REDIS_ERR;
- }
+ if (best == NULL) return REDIS_ERR;
key = dictGetEntryKey(best);
val = dictGetEntryVal(best);
lockThreadedIO();
if (listLength(server.io_newjobs) == 0) {
/* No new jobs in queue, exit. */
- redisLog(REDIS_DEBUG,"Thread %lld exiting, nothing to do",
- (long long) pthread_self());
+ redisLog(REDIS_DEBUG,"Thread %ld exiting, nothing to do",
+ (long) pthread_self());
server.io_active_threads--;
unlockThreadedIO();
return NULL;
listAddNodeTail(server.io_processing,j);
ln = listLast(server.io_processing); /* We use ln later to remove it */
unlockThreadedIO();
- redisLog(REDIS_DEBUG,"Thread %lld got a new job (type %d): %p about key '%s'",
- (long long) pthread_self(), j->type, (void*)j, (char*)j->key->ptr);
+ redisLog(REDIS_DEBUG,"Thread %ld got a new job (type %d): %p about key '%s'",
+ (long) pthread_self(), j->type, (void*)j, (char*)j->key->ptr);
/* Process the Job */
if (j->type == REDIS_IOJOB_LOAD) {
}
/* Done: insert the job into the processed queue */
- redisLog(REDIS_DEBUG,"Thread %lld completed the job: %p (key %s)",
- (long long) pthread_self(), (void*)j, (char*)j->key->ptr);
+ redisLog(REDIS_DEBUG,"Thread %ld completed the job: %p (key %s)",
+ (long) pthread_self(), (void*)j, (char*)j->key->ptr);
lockThreadedIO();
listDelNode(server.io_processing,ln);
listAddNodeTail(server.io_processed,j);
* Return 1 if the client is marked as blocked, 0 if the client can
* continue as the keys it is going to access appear to be in memory. */
static int blockClientOnSwappedKeys(struct redisCommand *cmd, redisClient *c) {
- if (cmd->proc == getCommand) {
- waitForSwappedKey(c,c->argv[1]);
- }
+ int j, last;
+
+ if (cmd->vm_firstkey == 0) return 0;
+ last = cmd->vm_lastkey;
+ if (last < 0) last = c->argc+last;
+ for (j = cmd->vm_firstkey; j <= last; j += cmd->vm_keystep)
+ waitForSwappedKey(c,c->argv[j]);
/* If the client was blocked for at least one key, mark it as blocked. */
if (listLength(c->io_keys)) {
c->flags |= REDIS_IO_WAIT;
}
key = dictGetEntryKey(de);
val = dictGetEntryVal(de);
- if (server.vm_enabled && (key->storage == REDIS_VM_MEMORY ||
- key->storage == REDIS_VM_SWAPPING)) {
+ if (!server.vm_enabled || (key->storage == REDIS_VM_MEMORY ||
+ key->storage == REDIS_VM_SWAPPING)) {
+ char *strenc;
+ char buf[128];
+
+ if (val->encoding < (sizeof(strencoding)/sizeof(char*))) {
+ strenc = strencoding[val->encoding];
+ } else {
+ snprintf(buf,64,"unknown encoding %d\n", val->encoding);
+ strenc = buf;
+ }
addReplySds(c,sdscatprintf(sdsempty(),
"+Key at:%p refcount:%d, value at:%p refcount:%d "
- "encoding:%d serializedlength:%lld\r\n",
+ "encoding:%s serializedlength:%lld\r\n",
(void*)key, key->refcount, (void*)val, val->refcount,
- val->encoding, (long long) rdbSavedObjectLen(val,NULL)));
+ strenc, (long long) rdbSavedObjectLen(val,NULL)));
} else {
addReplySds(c,sdscatprintf(sdsempty(),
"+Key at:%p refcount:%d, value swapped at: page %llu "
#else
return (void*) uc->uc_mcontext->__ss.__eip;
#endif
-#elif defined(__i386__) || defined(__X86_64__) || defined(__x86_64__)
+#elif defined(__i386__) || defined(__X86_64__) || defined(__x86_64__)
return (void*) uc->uc_mcontext.gregs[REG_EIP]; /* Linux 32/64 bit */
#elif defined(__ia64__) /* Linux IA64 */
return (void*) uc->uc_mcontext.sc_ip;
projects / redis.git / blobdiff